Convenient booking via chatbot

With our AI agent, we developed an innovative solution that enables room reservations via an intelligent chatbot. The chatbot can interact directly with users via text: it displays available rooms, suggests suitable time slots, answers questions, and immediately takes care of the booking—without any complicated forms. Instead of lots of clicks, users simply communicate with their digital assistant via chat or voice. This reduces complexity and takes the strain out of routine tasks.

The application is based on a large language model (LLM) that communicates with users in natural language and guides them through the entire reservation process. The LLM has access to various tools for interacting with the Kibanda API (e.g., reading available rooms, making reservations) and decides independently when to use which tool. The connection between the LLM and the Kibanda API is established via the Model Context Protocol (MCP). Published by Antrophic, this protocol defines how LLMs can retrieve and use available tools, prompts, and information, and is a de facto standard.

The solution was developed in Python and integrated directly into Microsoft Teams, Microsoft Copilot, and WhatsApp via Azure and Copilot Studio. This allows tenants to conveniently reserve rooms via their messenger—for example, with a simple message such as “I would like to reserve the laundry room for one hour tomorrow at 6 p.m.”—and receive an immediate response or booking confirmation.

In addition, the solution is designed so that it can also be connected to smart home systems such as Amazon Alexa or Google Home in the future to enable reservations via voice command, thus realizing fully networked, voice-controlled customer interaction.